Drug accumulation occurs in renal impairment, but there is currently a widespread erroneous view that metformin itself causes renal impairment (Box 6.1). Not many years ago, metformin would have been withdrawn when serum creatinine exceeded about 120 µmol/L (1.4 mg/dL), but this criterion had no evidence base, and it has become clear with wider use that adverse effects due to moderate renal impairment itself are extremely uncommon. The current consensus is that it should be withdrawn if estimated glomerular filtration rate (eGFR) is below 30 mL/min (e.g. serum creatinine < 200 µmol/L in a white male aged 60), but can be used safely, though with caution and at minimum effective dosage, if eGFR is 30–60 mL/min (60 mL/min is equivalent to serum creatinine of about 115 µmol/L (1.3 mg/dL) in the same patient).
The potency of metformin as an antihyperglycemic drug, especially in patients with renal impairment, should not be underestimated. The clinical impression is that HbA1c can increase by up to 3% (33 mmol/mol) if a substantial dose of metformin is abruptly withdrawn. It is the responsibility of the practitioner who withdraws metformin to ensure a plan is in place to manage the resulting, possibly severe, hyperglycemia. Transient deterioration in renal function in acutely ill hospitalized patients is very common, often precipitating abrupt withdrawal of metformin, but there is no reason for it not being restarted when renal function improves, though it often is not.
Severe hepatic impairment, and poorly controlled heart failure, are both considered contraindications (though many heart failure patients are treated with metformin, and 45% of them were taking metformin alone), but mildly abnormal liver function tests, very common in people with diabetes as a result of nonalcoholic fatty liver disease, are not a contraindication, and metformin (like the glitazones) may well cause a drop in transaminase levels. It is not contraindicated in patients who have had an uncomplicated myocardial infarction.
Lactic acidosis and vitamin B12 levels
Lactic acidosis usually occurs in the context of multiorgan failure, and it is often difficult to ascribe it to metformin use alone. Its incidence relative to metformin use overall is vanishingly small, and a systematic review concluded that there was no true increase in incidence in metformin-treated compared with non-metformin-treated diabetic patients . Vitamin B12 metabolism is subtly disturbed with long-term metformin treatment, but simple serum vitamin B12 levels have usually been normal. However, a large RCT over 4 years found that vitamin B12 deficiency (<150 pmol/L, 203 pg/mL) was 10% more common than in placebo-treated patients, with an associated increase in homocysteine levels. There was no increase in folate deficiency. Since B12 deficiency is simple to treat, there is now a strong case for measuring vitamin B12 levels every few years in patients taking long-term metformin treatment .
Radiological contrast medium-induced lactic acidosis
The conventional, but non-evidence-based, view is that metformin should be withheld 48 hours before and after a procedure involving an intravenous contrast agent, because of the increased risk of lactic acidosis (both agents are renally excreted unchanged). Cases nearly always occur in those with impaired renal function. Where possible, review previous laboratory results to establish trends in renal function, rather than using individual measurements.
- In patients with normal renal function (eGFR > 90 mL/min), continue metformin but ensure adequate hydration.
- Where there is renal impairment, withhold metformin for 48 hours after the procedure. Some patients will require alternative diabetes treatment over this period. Ensure adequate intravenous periprocedural hydration, and check renal function before restarting metformin.
Treatment failure with metformin: early use, or lifestyle first?
Measuring treatment failure with any agent is difficult, as the criteria for failure (often called ‘secondary failure’) have not been established. In ADOPT (2006) the failure rate for metformin treatment (defined as FPG > 10 mmol/L, 180 mg/dL) was only about 4% per year, similar to that with rosiglitazone, but lower than glibenclamide. However, in clinical practice the failure rate is much higher –17% per year judged by HbA1c exceeding 7.5% (59 mmol/mol) after treatment with metformin that reduced HbA1c to less than 7.0% (53 mmol/mol). Although many guidelines continue to propose diet and lifestyle first, progressing to metformin treatment only if the non-drug strategy fails (e.g. NICE 2009), where acceptable to patients metformin should be offered at diagnosis, together with intensive lifestyle intervention on the basis that lifestyle effects are likely to wane after 3–6 months in the real-life setting .
Sulfonylureas and meglitinides (prandial insulin regulators) (BNF, section 184.108.40.206)
Sulphonylureas have been workhorses of diabetes therapy for over 40 years. The short-acting meglitinides, operating through similar mechanisms, are much more recent. Neither should be used as first-line treatment in normal weight, overweight or obese patients, or in those with multiple insulin resistance characteristics, where first-line metformin is preferable (Table 6.1).
Sulfonylureas stimulate pancreatic insulin secretion by binding to the sulphonylurea receptor (SUR1) on the J3 cell. This closes Kir 6.2 (K-ATP) channels, resulting in calcium influx, stimulating release initially of preformed insulin granules (first-phase insulin secretion), and thereafter increasing release and formation of insulin-containing granules (second-phase insulin secretion). It is the prolonged stimulation of insulin release, independent of ambient glucose levels, that increases the risk of profound and long-lasting hypoglycemia with these drugs. The only truly short- acting agent, tolbutamide, is no longer widely used (though apart from the necessary multiple daily dosing, its demise has been largely dictated by general usage and fashion) and the distinction between the remaining members of the class as either intermediate or long-acting is pharmacokinetic and does not reliably characterize their clinical duration of action, nor the likelihood of their causing severe hypoglycemia. They are highly protein-bound and can interact with warfarin and salicylates, increasing the risk of hypoglycemia.
Potential adverse cardiovascular effects
Some sulphonylureas, especially glibenclamide, cross-react with and close cardiac potassium channels and experimentally reduce protective cardiac ischemic preconditioning. Whether this effect is clinically relevant is not clear, but it is not likely to be a major consideration. Gliclazide, the sulphonylurea most used in the UK, does not have this potential cardiac effect; neither does glimepiride, nor the meglitinides. Though there are few long-term studies, RCT evidence is reassuring: in ADOPT (2006), newly diagnosed patients treated with glibenclamide up to 15 mg daily had a lower cardiac event rate than those treated with metformin or rosiglitazone monotherapy, and there was no excess of cardiovascular events in UKPDS with either glibenclamide or the very long-acting chlorpropamide, no longer used.